Bug 346789

Comment 1 Marc Khouzam 2011-05-21 11:37:39 EDT

Created attachment 196271 [details]
First part of fix

The problem was that we now clear the launch attributes before every class of tests, therefore when a suite sets an attribute, that attribute is lost as soon as the first class of test is loaded.

This patch adds a concept of globalLaunchAttributes which will not be reset when loading a new class of tests.  Those global attributes can be used by suites, but the suite must clear them once it is done.

Committed to HEAD

Comment 2 CDT Genie 2011-05-21 12:23:08 EDT

*** cdt cvs genie on behalf of mkhouzam ***
Bug 346789: Remote JUnits don't work anymore

[*] Suite_Remote_7_2.java 1.6 http://dev.eclipse.org/viewcvs/index.cgi/org.eclipse.cdt/dsf-gdb/org.eclipse.cdt.tests.dsf.gdb/src/org/eclipse/cdt/tests/dsf/gdb/tests/tests_7_2/Suite_Remote_7_2.java?root=Tools_Project&r1=1.5&r2=1.6

[*] Suite_Remote_7_1.java 1.6 http://dev.eclipse.org/viewcvs/index.cgi/org.eclipse.cdt/dsf-gdb/org.eclipse.cdt.tests.dsf.gdb/src/org/eclipse/cdt/tests/dsf/gdb/tests/tests_7_1/Suite_Remote_7_1.java?root=Tools_Project&r1=1.5&r2=1.6

[*] Suite_Remote_7_0.java 1.8 http://dev.eclipse.org/viewcvs/index.cgi/org.eclipse.cdt/dsf-gdb/org.eclipse.cdt.tests.dsf.gdb/src/org/eclipse/cdt/tests/dsf/gdb/tests/tests_7_0/Suite_Remote_7_0.java?root=Tools_Project&r1=1.7&r2=1.8

[*] Suite_Remote_7_3.java 1.3 http://dev.eclipse.org/viewcvs/index.cgi/org.eclipse.cdt/dsf-gdb/org.eclipse.cdt.tests.dsf.gdb/src/org/eclipse/cdt/tests/dsf/gdb/tests/tests_7_3/Suite_Remote_7_3.java?root=Tools_Project&r1=1.2&r2=1.3

[*] Suite_Remote_6_6.java 1.6 http://dev.eclipse.org/viewcvs/index.cgi/org.eclipse.cdt/dsf-gdb/org.eclipse.cdt.tests.dsf.gdb/src/org/eclipse/cdt/tests/dsf/gdb/tests/tests_6_6/Suite_Remote_6_6.java?root=Tools_Project&r1=1.5&r2=1.6

[*] Suite_Remote_6_8.java 1.6 http://dev.eclipse.org/viewcvs/index.cgi/org.eclipse.cdt/dsf-gdb/org.eclipse.cdt.tests.dsf.gdb/src/org/eclipse/cdt/tests/dsf/gdb/tests/tests_6_8/Suite_Remote_6_8.java?root=Tools_Project&r1=1.5&r2=1.6

[*] BaseTestCase.java 1.17 http://dev.eclipse.org/viewcvs/index.cgi/org.eclipse.cdt/dsf-gdb/org.eclipse.cdt.tests.dsf.gdb/src/org/eclipse/cdt/tests/dsf/gdb/framework/BaseTestCase.java?root=Tools_Project&r1=1.16&r2=1.17
[+] BaseRemoteSuite.java  http://dev.eclipse.org/viewcvs/index.cgi/org.eclipse.cdt/dsf-gdb/org.eclipse.cdt.tests.dsf.gdb/src/org/eclipse/cdt/tests/dsf/gdb/framework/BaseRemoteSuite.java?root=Tools_Project&revision=1.1&view=markup

[*] Suite_Remote_6_7.java 1.6 http://dev.eclipse.org/viewcvs/index.cgi/org.eclipse.cdt/dsf-gdb/org.eclipse.cdt.tests.dsf.gdb/src/org/eclipse/cdt/tests/dsf/gdb/tests/tests_6_7/Suite_Remote_6_7.java?root=Tools_Project&r1=1.5&r2=1.6

Comment 3 Marc Khouzam 2012-02-02 06:38:57 EST

Note that there was a memory leak that prevented us from running more than 140 remote tests.  This has been fixed by Bug 370376

Comment 4 Marc Khouzam 2012-02-09 15:46:49 EST

Created attachment 210823 [details]
Properly detect that remote tests have stopped at main

There is a problem with the startup sequence of the remote tests.

The tests need to know when we are properly stopped at main.

With GDB >= 7.0, when connecting to the remote gdbserver, we get a first *stopped event at connection time.  This is not the event we want.  I had modified the code to instead listen for an IBreakpointHitDMEvent.

However, with GDB <= 6.8, temporary breakpoints are not reported as breakpoint-hit, so we don't get an IBreakpointHitDMEvent for GDB <= 6.8 but simply an ISuspendedDMEvent.

So, the question was, how can we listen for an ISuspendedDMEvent, but know that for GDB >= 7.0 we should ignore the first one?

What I found to be able to know we have stopped at main, in all cases,
is to look for an ISuspendedDMEvent and then confirming that it indicates
in its frame that it stopped at "main".  This will allow us to skip
the first *stopped event for GDB >= 7.0 and also properly detect the stop for GDB <= 6.8

The attach patch does this.
2007 extra remote tests now pass :)

I will commit to master.

Comment 5 Marc Khouzam 2012-02-10 05:49:02 EST

Remote tests now work overall.  Some individual failures need to be fixed, but I'll address that in a separate bug.

Comment 6 CDT Genie 2012-02-29 13:24:47 EST

*** cdt git genie on behalf of Marc Khouzam ***

    Bug 346789: fix startup sequence for remote JUnits tests

[*] http://git.eclipse.org/c/cdt/org.eclipse.cdt.git/commit/?id=7f07eb597c9255655efd9895093ac7fb8ac42127

Comment 7 CDT Genie 2012-02-29 13:24:48 EST

*** cdt git genie on behalf of Marc Khouzam ***

    Bug 346789: fix startup sequence for remote JUnits tests

[*] http://git.eclipse.org/c/cdt/org.eclipse.cdt.git/commit/?id=7f07eb597c9255655efd9895093ac7fb8ac42127
*** cdt git genie on behalf of Marc Khouzam ***

    Bug 346789: fix startup sequence for remote JUnits tests

[*] http://git.eclipse.org/c/cdt/org.eclipse.cdt.git/commit/?id=7f07eb597c9255655efd9895093ac7fb8ac42127